NASA Space Apps · Planet X and SPHEREx
Astroswipe turns two flagship surveys into a crowdsourced blink comparator, so anyone can help find what moves or changes in the infrared sky.
Launched in 2025, SPHEREx maps the entire sky every ~6 months in 102 near-infrared bands from 0.75 to 5.0 µm. Each survey is a new epoch: comparing them reveals objects that move, brighten or fade, and the spectra show what they're made of.
Gaia measured positions, parallaxes and proper motions for ~1.8 billion stars. Cross-matching with Gaia DR3 removes known movers, so what remains in SPHEREx difference images is genuinely uncatalogued.
Flipping between photographic plates on a mechanical blink comparator at Lowell Observatory.
Revealing cold brown dwarfs like WISE 0855, the coldest known, at ~250 K.
Volunteers blinking WISE images find thousands of brown dwarf candidates.
Every 6 months a fresh 102-band all-sky map — reviewed by swiping, one patch at a time.
Extremely cold, slow movers: a faint dipole that shifts a few pixels between epochs.
Bright at 4–5 µm, faint below 1 µm, with methane and water absorption, often with high proper motion.
Stay put but change flux: a single-signed residual in the difference image.
Switch between SPHEREx's six detector arrays or a false-colour composite spanning 0.75–5.0 µm.
Try it →Epoch subtraction turns motion into black/white dipoles and brightening into residual flux.
Try it →Known Gaia sources are circled with motion vectors so uncatalogued movers stand out.
Try it →Every patch renders a 102-point spectral energy distribution in the Deep Dive inspector.
Try it →Zoom from wide survey context down to arcminute scale on any card.
Try it →Any view is shareable and embeddable via ?ra=…&dec=…&fov=… parameters.
Try it →Each SPHEREx epoch is a huge volume of data, far too much for a few astronomers to inspect by eye. Astroswipe splits it into millions of small, deterministic sky patches that take a few seconds each to judge.
Volunteers mark a patch static or "something changed".
Flags go to 10 independent volunteers; 70% agreement confirms.
Experienced volunteers settle disputed cases.
Vetted candidates export as structured reports for follow-up.
Server-side duplicate-vote protection and rate limits keep consensus honest, so astronomers get a short list of high-confidence candidates instead of raw survey data.